fix: percent-encode spaces in request path URLs

[sylvesterkaczmarek]

Summary

Make HttpRequest.url() encode spaces in path segments as %20 instead of +, matching URL path semantics and the URL actually constructed by the OkHttp transport.

Fixes #886.

Problem

HttpRequest.url() currently uses URLEncoder for both path segments and query parameters.

For query parameters, form encoding is appropriate and a space becomes +.

For path segments, however, + is a literal plus character. A request containing:

.addPathSegment("user name")

currently renders as:

https://api.example.com/user+name

The OkHttp transport itself constructs the request with HttpUrl.Builder.addPathSegment, so the actual wire URL uses:

https://api.example.com/user%20name

LoggingHttpClient prints request.url(), which means SDK logs can show a different request target from the one actually sent.

Fix

Keep the existing URLEncoder behavior for query components, but normalize the encoded path-segment space representation:

URLEncoder.encode(segment, "UTF-8").replace("+", "%20")

A literal plus remains unambiguous because URLEncoder already encodes it as %2B before the replacement is applied.

Regression coverage

Updated HttpRequestTest to verify:

• a path segment containing a space renders with %20;
• a literal + path character renders as %2B;
• the existing query-space case remains hello+world.

Validation

• branch is based directly on current upstream main at cf942a40074291290634321ad9fe21e514030b4c;
• branch is 0 commits behind upstream;
• production diff is exactly 1 addition / 1 deletion in HttpRequest.kt;
• the actual OkHttp request-construction path is unchanged.

Full repository validation is left to GitHub Actions.

Risk

Low. This changes only the diagnostic/string rendering of path-segment spaces. Query encoding and transport request construction are unchanged.